Training in progress, step 8000, checkpoint

checkpoint-8000/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8760b1447a3094a149e7c4bd1380bbfdf581ca4220024fcd0b8739bad683240b
+oid sha256:fd18949200cc18253784c7a7a9eb2edd3a1bd2b28eaae64ac118b150d9c5cab1
 size 577789320

checkpoint-8000/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ccc6f19758766e01b7282857274d15516c4fc7f623633cb56154c016270ecb57
+oid sha256:2b2d534f5a1509cc19619065899e388a3e7638e5ffb644652de86750bb93f12a
 size 1155772233

checkpoint-8000/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8840c2422ad88a77f460b571a22c90bffa453c8f4f191f77fe10505224787750
+oid sha256:f32741e19bb28c33b69dd067716472e7e4f838444bd374bcfd9b816ca07aa38a
 size 14244

checkpoint-8000/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0aef1303f298e810b9c563fac5650fcaebc5d5e2a568db30104db374f503c826
+oid sha256:03842876983aebd886d0e95b5f18afd54f0a345fd4849cda48bff8e02fa53f4a
 size 1064

checkpoint-8000/trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.4122372269630432,
-  "best_model_checkpoint": "mikhail-panzo/zlm-fil-ceb_b64_le5_s8000/checkpoint-5000",
-  "epoch": 347.82608695652175,
+  "best_metric": 0.3926040828227997,
+  "best_model_checkpoint": "mikhail-panzo/zlm-fil-ceb_b64_le5_s8000/checkpoint-6500",
+  "epoch": 316.83168316831683,
   "eval_steps": 500,
   "global_step": 8000,
   "is_hyper_param_search": false,
@@ -9,1258 +9,1258 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 2.1739130434782608,
-      "grad_norm": 0.7357823252677917,
+      "epoch": 1.9801980198019802,
+      "grad_norm": 1.2683665752410889,
       "learning_rate": 2.5000000000000004e-07,
-      "loss": 0.4055,
+      "loss": 0.4932,
       "step": 50
     },
     {
-      "epoch": 4.3478260869565215,
-      "grad_norm": 0.7944665551185608,
+      "epoch": 3.9603960396039604,
+      "grad_norm": 1.1266463994979858,
       "learning_rate": 5.000000000000001e-07,
-      "loss": 0.3988,
+      "loss": 0.4898,
       "step": 100
     },
     {
-      "epoch": 6.521739130434782,
-      "grad_norm": 0.7226569652557373,
+      "epoch": 5.9405940594059405,
+      "grad_norm": 1.1800968647003174,
       "learning_rate": 7.5e-07,
-      "loss": 0.4072,
+      "loss": 0.4839,
       "step": 150
     },
     {
-      "epoch": 8.695652173913043,
-      "grad_norm": 0.7845238447189331,
+      "epoch": 7.920792079207921,
+      "grad_norm": 0.8235568404197693,
       "learning_rate": 1.0000000000000002e-06,
-      "loss": 0.4028,
+      "loss": 0.4785,
       "step": 200
     },
     {
-      "epoch": 10.869565217391305,
-      "grad_norm": 0.9769566655158997,
+      "epoch": 9.900990099009901,
+      "grad_norm": 1.313211441040039,
       "learning_rate": 1.25e-06,
-      "loss": 0.4066,
+      "loss": 0.4767,
       "step": 250
     },
     {
-      "epoch": 13.043478260869565,
-      "grad_norm": 0.8634074330329895,
+      "epoch": 11.881188118811881,
+      "grad_norm": 0.7831560373306274,
       "learning_rate": 1.5e-06,
-      "loss": 0.405,
+      "loss": 0.4681,
       "step": 300
     },
     {
-      "epoch": 15.217391304347826,
-      "grad_norm": 0.9029327034950256,
+      "epoch": 13.861386138613861,
+      "grad_norm": 0.7987237572669983,
       "learning_rate": 1.75e-06,
-      "loss": 0.4074,
+      "loss": 0.4658,
       "step": 350
     },
     {
-      "epoch": 17.391304347826086,
-      "grad_norm": 0.7624074220657349,
+      "epoch": 15.841584158415841,
+      "grad_norm": 0.7143813371658325,
       "learning_rate": 2.0000000000000003e-06,
-      "loss": 0.406,
+      "loss": 0.4627,
       "step": 400
     },
     {
-      "epoch": 19.565217391304348,
-      "grad_norm": 1.005196213722229,
+      "epoch": 17.821782178217823,
+      "grad_norm": 0.8037531971931458,
       "learning_rate": 2.25e-06,
-      "loss": 0.4052,
+      "loss": 0.461,
       "step": 450
     },
     {
-      "epoch": 21.73913043478261,
-      "grad_norm": 1.0276380777359009,
+      "epoch": 19.801980198019802,
+      "grad_norm": 0.79031902551651,
       "learning_rate": 2.5e-06,
-      "loss": 0.4019,
+      "loss": 0.4592,
       "step": 500
     },
     {
-      "epoch": 21.73913043478261,
-      "eval_loss": 0.4143177568912506,
-      "eval_runtime": 6.4074,
-      "eval_samples_per_second": 24.815,
-      "eval_steps_per_second": 3.121,
+      "epoch": 19.801980198019802,
+      "eval_loss": 0.42528408765792847,
+      "eval_runtime": 7.8982,
+      "eval_samples_per_second": 22.79,
+      "eval_steps_per_second": 2.912,
       "step": 500
     },
     {
-      "epoch": 23.91304347826087,
-      "grad_norm": 0.9067140817642212,
+      "epoch": 21.782178217821784,
+      "grad_norm": 0.7966288328170776,
       "learning_rate": 2.7500000000000004e-06,
-      "loss": 0.4021,
+      "loss": 0.4551,
       "step": 550
     },
     {
-      "epoch": 26.08695652173913,
-      "grad_norm": 0.8799042701721191,
+      "epoch": 23.762376237623762,
+      "grad_norm": 0.8126183152198792,
       "learning_rate": 3e-06,
-      "loss": 0.4071,
+      "loss": 0.4542,
       "step": 600
     },
     {
-      "epoch": 28.26086956521739,
-      "grad_norm": 1.0451533794403076,
+      "epoch": 25.742574257425744,
+      "grad_norm": 0.6602805852890015,
       "learning_rate": 3.2500000000000002e-06,
-      "loss": 0.4146,
+      "loss": 0.4535,
       "step": 650
     },
     {
-      "epoch": 30.434782608695652,
-      "grad_norm": 0.7998800277709961,
+      "epoch": 27.722772277227723,
+      "grad_norm": 0.7177000641822815,
       "learning_rate": 3.5e-06,
-      "loss": 0.4023,
+      "loss": 0.4484,
       "step": 700
     },
     {
-      "epoch": 32.608695652173914,
-      "grad_norm": 1.0301792621612549,
+      "epoch": 29.702970297029704,
+      "grad_norm": 0.7912996411323547,
       "learning_rate": 3.7500000000000005e-06,
-      "loss": 0.4045,
+      "loss": 0.4486,
       "step": 750
     },
     {
-      "epoch": 34.78260869565217,
-      "grad_norm": 0.6388185620307922,
+      "epoch": 31.683168316831683,
+      "grad_norm": 0.6251747608184814,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.4045,
+      "loss": 0.4463,
       "step": 800
     },
     {
-      "epoch": 36.95652173913044,
-      "grad_norm": 0.9676840901374817,
+      "epoch": 33.663366336633665,
+      "grad_norm": 1.003780722618103,
       "learning_rate": 4.25e-06,
-      "loss": 0.4116,
+      "loss": 0.4474,
       "step": 850
     },
     {
-      "epoch": 39.130434782608695,
-      "grad_norm": 0.8355514407157898,
+      "epoch": 35.64356435643565,
+      "grad_norm": 0.742171585559845,
       "learning_rate": 4.5e-06,
-      "loss": 0.4013,
+      "loss": 0.4434,
       "step": 900
     },
     {
-      "epoch": 41.30434782608695,
-      "grad_norm": 0.7534067630767822,
+      "epoch": 37.62376237623762,
+      "grad_norm": 0.6034647822380066,
       "learning_rate": 4.75e-06,
-      "loss": 0.4019,
+      "loss": 0.4425,
       "step": 950
     },
     {
-      "epoch": 43.47826086956522,
-      "grad_norm": 0.9366117119789124,
+      "epoch": 39.603960396039604,
+      "grad_norm": 0.6990635395050049,
       "learning_rate": 5e-06,
-      "loss": 0.4064,
+      "loss": 0.4381,
       "step": 1000
     },
     {
-      "epoch": 43.47826086956522,
-      "eval_loss": 0.41316837072372437,
-      "eval_runtime": 6.5616,
-      "eval_samples_per_second": 24.232,
-      "eval_steps_per_second": 3.048,
+      "epoch": 39.603960396039604,
+      "eval_loss": 0.41002073884010315,
+      "eval_runtime": 6.7782,
+      "eval_samples_per_second": 26.556,
+      "eval_steps_per_second": 3.393,
       "step": 1000
     },
     {
-      "epoch": 45.65217391304348,
-      "grad_norm": 0.7025179266929626,
+      "epoch": 41.584158415841586,
+      "grad_norm": 0.9420008659362793,
       "learning_rate": 5.2500000000000006e-06,
-      "loss": 0.4061,
+      "loss": 0.4421,
       "step": 1050
     },
     {
-      "epoch": 47.82608695652174,
-      "grad_norm": 0.9240646958351135,
+      "epoch": 43.56435643564357,
+      "grad_norm": 0.7143107652664185,
       "learning_rate": 5.500000000000001e-06,
-      "loss": 0.4061,
+      "loss": 0.4389,
       "step": 1100
     },
     {
-      "epoch": 50.0,
-      "grad_norm": 0.9685391187667847,
+      "epoch": 45.54455445544554,
+      "grad_norm": 0.7934616804122925,
       "learning_rate": 5.75e-06,
-      "loss": 0.4063,
+      "loss": 0.4369,
       "step": 1150
     },
     {
-      "epoch": 52.17391304347826,
-      "grad_norm": 0.8591094017028809,
+      "epoch": 47.524752475247524,
+      "grad_norm": 0.9682563543319702,
       "learning_rate": 6e-06,
-      "loss": 0.4013,
+      "loss": 0.4355,
       "step": 1200
     },
     {
-      "epoch": 54.34782608695652,
-      "grad_norm": 0.9348228573799133,
+      "epoch": 49.504950495049506,
+      "grad_norm": 0.7880070209503174,
       "learning_rate": 6.25e-06,
-      "loss": 0.402,
+      "loss": 0.4359,
       "step": 1250
     },
     {
-      "epoch": 56.52173913043478,
-      "grad_norm": 0.7908409237861633,
+      "epoch": 51.48514851485149,
+      "grad_norm": 0.8443828821182251,
       "learning_rate": 6.5000000000000004e-06,
-      "loss": 0.4046,
+      "loss": 0.4324,
       "step": 1300
     },
     {
-      "epoch": 58.69565217391305,
-      "grad_norm": 1.109017014503479,
+      "epoch": 53.46534653465346,
+      "grad_norm": 0.724743127822876,
       "learning_rate": 6.750000000000001e-06,
-      "loss": 0.3995,
+      "loss": 0.4348,
       "step": 1350
     },
     {
-      "epoch": 60.869565217391305,
-      "grad_norm": 0.8529478311538696,
+      "epoch": 55.445544554455445,
+      "grad_norm": 0.7557791471481323,
       "learning_rate": 7e-06,
-      "loss": 0.4018,
+      "loss": 0.433,
       "step": 1400
     },
     {
-      "epoch": 63.04347826086956,
-      "grad_norm": 0.9947476387023926,
+      "epoch": 57.42574257425743,
+      "grad_norm": 0.7936097383499146,
       "learning_rate": 7.25e-06,
-      "loss": 0.401,
+      "loss": 0.4273,
       "step": 1450
     },
     {
-      "epoch": 65.21739130434783,
-      "grad_norm": 1.2476813793182373,
+      "epoch": 59.40594059405941,
+      "grad_norm": 0.7041624784469604,
       "learning_rate": 7.500000000000001e-06,
-      "loss": 0.4034,
+      "loss": 0.4281,
       "step": 1500
     },
     {
-      "epoch": 65.21739130434783,
-      "eval_loss": 0.41223829984664917,
-      "eval_runtime": 6.5846,
-      "eval_samples_per_second": 24.147,
-      "eval_steps_per_second": 3.037,
+      "epoch": 59.40594059405941,
+      "eval_loss": 0.40218469500541687,
+      "eval_runtime": 7.4309,
+      "eval_samples_per_second": 24.223,
+      "eval_steps_per_second": 3.095,
       "step": 1500
     },
     {
-      "epoch": 67.3913043478261,
-      "grad_norm": 0.8658013343811035,
+      "epoch": 61.386138613861384,
+      "grad_norm": 0.9194286465644836,
       "learning_rate": 7.75e-06,
-      "loss": 0.4007,
+      "loss": 0.4275,
       "step": 1550
     },
     {
-      "epoch": 69.56521739130434,
-      "grad_norm": 0.7932788133621216,
+      "epoch": 63.366336633663366,
+      "grad_norm": 0.7239477634429932,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.4054,
+      "loss": 0.4245,
       "step": 1600
     },
     {
-      "epoch": 71.73913043478261,
-      "grad_norm": 2.33347225189209,
+      "epoch": 65.34653465346534,
+      "grad_norm": 0.7235706448554993,
       "learning_rate": 8.25e-06,
-      "loss": 0.4075,
+      "loss": 0.4248,
       "step": 1650
     },
     {
-      "epoch": 73.91304347826087,
-      "grad_norm": 1.0095164775848389,
+      "epoch": 67.32673267326733,
+      "grad_norm": 0.7281640768051147,
       "learning_rate": 8.5e-06,
-      "loss": 0.3982,
+      "loss": 0.425,
       "step": 1700
     },
     {
-      "epoch": 76.08695652173913,
-      "grad_norm": 0.8895902633666992,
+      "epoch": 69.3069306930693,
+      "grad_norm": 1.011294960975647,
       "learning_rate": 8.750000000000001e-06,
-      "loss": 0.4018,
+      "loss": 0.4238,
       "step": 1750
     },
     {
-      "epoch": 78.26086956521739,
-      "grad_norm": 1.2252330780029297,
+      "epoch": 71.2871287128713,
+      "grad_norm": 1.1249107122421265,
       "learning_rate": 9e-06,
-      "loss": 0.4024,
+      "loss": 0.4203,
       "step": 1800
     },
     {
-      "epoch": 80.43478260869566,
-      "grad_norm": 0.7731898427009583,
+      "epoch": 73.26732673267327,
+      "grad_norm": 0.7369679808616638,
       "learning_rate": 9.250000000000001e-06,
-      "loss": 0.3994,
+      "loss": 0.422,
       "step": 1850
     },
     {
-      "epoch": 82.6086956521739,
-      "grad_norm": 1.047607660293579,
+      "epoch": 75.24752475247524,
+      "grad_norm": 0.704367995262146,
       "learning_rate": 9.5e-06,
-      "loss": 0.4013,
+      "loss": 0.4201,
       "step": 1900
     },
     {
-      "epoch": 84.78260869565217,
-      "grad_norm": 0.9284490346908569,
+      "epoch": 77.22772277227723,
+      "grad_norm": 0.8307437896728516,
       "learning_rate": 9.75e-06,
-      "loss": 0.3986,
+      "loss": 0.4203,
       "step": 1950
     },
     {
-      "epoch": 86.95652173913044,
-      "grad_norm": 4.725491046905518,
+      "epoch": 79.20792079207921,
+      "grad_norm": 0.7827133536338806,
       "learning_rate": 1e-05,
-      "loss": 0.3985,
+      "loss": 0.4195,
       "step": 2000
     },
     {
-      "epoch": 86.95652173913044,
-      "eval_loss": 0.4147759974002838,
-      "eval_runtime": 6.5475,
-      "eval_samples_per_second": 24.284,
-      "eval_steps_per_second": 3.055,
+      "epoch": 79.20792079207921,
+      "eval_loss": 0.39878609776496887,
+      "eval_runtime": 8.1579,
+      "eval_samples_per_second": 22.064,
+      "eval_steps_per_second": 2.819,
       "step": 2000
     },
     {
-      "epoch": 89.1304347826087,
-      "grad_norm": 1.1678683757781982,
+      "epoch": 81.18811881188118,
+      "grad_norm": 0.8909313082695007,
       "learning_rate": 9.916666666666668e-06,
-      "loss": 0.4066,
+      "loss": 0.4203,
       "step": 2050
     },
     {
-      "epoch": 91.30434782608695,
-      "grad_norm": 1.988890528678894,
-      "learning_rate": 9.835000000000002e-06,
-      "loss": 0.3997,
+      "epoch": 83.16831683168317,
+      "grad_norm": 0.7147483825683594,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 0.418,
       "step": 2100
     },
     {
-      "epoch": 93.47826086956522,
-      "grad_norm": 1.0278427600860596,
-      "learning_rate": 9.751666666666667e-06,
-      "loss": 0.3948,
+      "epoch": 85.14851485148515,
+      "grad_norm": 0.6844545006752014,
+      "learning_rate": 9.75e-06,
+      "loss": 0.4167,
       "step": 2150
     },
     {
-      "epoch": 95.65217391304348,
-      "grad_norm": 0.9606735110282898,
-      "learning_rate": 9.67e-06,
-      "loss": 0.42,
+      "epoch": 87.12871287128714,
+      "grad_norm": 0.88512122631073,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 0.4108,
       "step": 2200
     },
     {
-      "epoch": 97.82608695652173,
-      "grad_norm": 1.0861903429031372,
-      "learning_rate": 9.586666666666667e-06,
-      "loss": 0.398,
+      "epoch": 89.10891089108911,
+      "grad_norm": 0.7097252607345581,
+      "learning_rate": 9.583333333333335e-06,
+      "loss": 0.4135,
       "step": 2250
     },
     {
-      "epoch": 100.0,
-      "grad_norm": 1.4872643947601318,
-      "learning_rate": 9.503333333333334e-06,
-      "loss": 0.4059,
+      "epoch": 91.08910891089108,
+      "grad_norm": 0.8400272727012634,
+      "learning_rate": 9.5e-06,
+      "loss": 0.4147,
       "step": 2300
     },
     {
-      "epoch": 102.17391304347827,
-      "grad_norm": 1.0460981130599976,
-      "learning_rate": 9.42e-06,
-      "loss": 0.4014,
+      "epoch": 93.06930693069307,
+      "grad_norm": 0.7614939212799072,
+      "learning_rate": 9.416666666666667e-06,
+      "loss": 0.4119,
       "step": 2350
     },
     {
-      "epoch": 104.34782608695652,
-      "grad_norm": 1.906476378440857,
-      "learning_rate": 9.336666666666666e-06,
-      "loss": 0.4015,
+      "epoch": 95.04950495049505,
+      "grad_norm": 0.8099732995033264,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 0.4108,
       "step": 2400
     },
     {
-      "epoch": 106.52173913043478,
-      "grad_norm": 1.3965051174163818,
-      "learning_rate": 9.253333333333333e-06,
-      "loss": 0.4063,
+      "epoch": 97.02970297029702,
+      "grad_norm": 1.0697815418243408,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 0.4095,
       "step": 2450
     },
     {
-      "epoch": 108.69565217391305,
-      "grad_norm": 1.7410107851028442,
-      "learning_rate": 9.17e-06,
-      "loss": 0.396,
+      "epoch": 99.00990099009901,
+      "grad_norm": 0.7235235571861267,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 0.4134,
       "step": 2500
     },
     {
-      "epoch": 108.69565217391305,
-      "eval_loss": 0.41233959794044495,
-      "eval_runtime": 6.6031,
-      "eval_samples_per_second": 24.08,
-      "eval_steps_per_second": 3.029,
+      "epoch": 99.00990099009901,
+      "eval_loss": 0.39545077085494995,
+      "eval_runtime": 7.724,
+      "eval_samples_per_second": 23.304,
+      "eval_steps_per_second": 2.978,
       "step": 2500
     },
     {
-      "epoch": 110.8695652173913,
-      "grad_norm": 1.4562053680419922,
-      "learning_rate": 9.088333333333334e-06,
-      "loss": 0.4042,
+      "epoch": 100.99009900990099,
+      "grad_norm": 0.6166990399360657,
+      "learning_rate": 9.083333333333333e-06,
+      "loss": 0.4097,
       "step": 2550
     },
     {
-      "epoch": 113.04347826086956,
-      "grad_norm": 0.8741047978401184,
-      "learning_rate": 9.005000000000001e-06,
-      "loss": 0.3985,
+      "epoch": 102.97029702970298,
+      "grad_norm": 0.6944295763969421,
+      "learning_rate": 9e-06,
+      "loss": 0.4107,
       "step": 2600
     },
     {
-      "epoch": 115.21739130434783,
-      "grad_norm": 1.1817028522491455,
-      "learning_rate": 8.921666666666668e-06,
-      "loss": 0.4036,
+      "epoch": 104.95049504950495,
+      "grad_norm": 0.8469858765602112,
+      "learning_rate": 8.916666666666667e-06,
+      "loss": 0.4056,
       "step": 2650
     },
     {
-      "epoch": 117.3913043478261,
-      "grad_norm": 0.8515534996986389,
-      "learning_rate": 8.838333333333335e-06,
-      "loss": 0.3986,
+      "epoch": 106.93069306930693,
+      "grad_norm": 0.6261239647865295,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 0.409,
       "step": 2700
     },
     {
-      "epoch": 119.56521739130434,
-      "grad_norm": 0.9545428156852722,
-      "learning_rate": 8.755e-06,
-      "loss": 0.3946,
+      "epoch": 108.91089108910892,
+      "grad_norm": 0.6304221749305725,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.4081,
       "step": 2750
     },
     {
-      "epoch": 121.73913043478261,
-      "grad_norm": 0.6653321385383606,
-      "learning_rate": 8.671666666666667e-06,
-      "loss": 0.3957,
+      "epoch": 110.89108910891089,
+      "grad_norm": 0.6513387560844421,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 0.4071,
       "step": 2800
     },
     {
-      "epoch": 123.91304347826087,
-      "grad_norm": 0.8100114464759827,
-      "learning_rate": 8.588333333333334e-06,
-      "loss": 0.3953,
+      "epoch": 112.87128712871286,
+      "grad_norm": 0.9311261177062988,
+      "learning_rate": 8.583333333333333e-06,
+      "loss": 0.4091,
       "step": 2850
     },
     {
-      "epoch": 126.08695652173913,
-      "grad_norm": 1.4174237251281738,
-      "learning_rate": 8.505e-06,
-      "loss": 0.3933,
+      "epoch": 114.85148514851485,
+      "grad_norm": 0.7296612858772278,
+      "learning_rate": 8.5e-06,
+      "loss": 0.4055,
       "step": 2900
     },
     {
-      "epoch": 128.2608695652174,
-      "grad_norm": 1.6101123094558716,
-      "learning_rate": 8.421666666666668e-06,
-      "loss": 0.4027,
+      "epoch": 116.83168316831683,
+      "grad_norm": 0.8149676322937012,
+      "learning_rate": 8.416666666666667e-06,
+      "loss": 0.4052,
       "step": 2950
     },
     {
-      "epoch": 130.43478260869566,
-      "grad_norm": 0.7453967332839966,
-      "learning_rate": 8.338333333333335e-06,
-      "loss": 0.3958,
+      "epoch": 118.81188118811882,
+      "grad_norm": 0.8266366124153137,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.4049,
       "step": 3000
     },
     {
-      "epoch": 130.43478260869566,
-      "eval_loss": 0.4138609766960144,
-      "eval_runtime": 6.6093,
-      "eval_samples_per_second": 24.057,
-      "eval_steps_per_second": 3.026,
+      "epoch": 118.81188118811882,
+      "eval_loss": 0.39345207810401917,
+      "eval_runtime": 7.2151,
+      "eval_samples_per_second": 24.948,
+      "eval_steps_per_second": 3.188,
       "step": 3000
     },
     {
-      "epoch": 132.6086956521739,
-      "grad_norm": 1.061684012413025,
-      "learning_rate": 8.255000000000001e-06,
-      "loss": 0.3966,
+      "epoch": 120.79207920792079,
+      "grad_norm": 0.6681864261627197,
+      "learning_rate": 8.25e-06,
+      "loss": 0.4022,
       "step": 3050
     },
     {
-      "epoch": 134.7826086956522,
-      "grad_norm": 1.0820105075836182,
-      "learning_rate": 8.171666666666668e-06,
-      "loss": 0.3937,
+      "epoch": 122.77227722772277,
+      "grad_norm": 0.5798916220664978,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 0.4026,
       "step": 3100
     },
     {
-      "epoch": 136.95652173913044,
-      "grad_norm": 0.8147062063217163,
-      "learning_rate": 8.088333333333334e-06,
-      "loss": 0.3961,
+      "epoch": 124.75247524752476,
+      "grad_norm": 0.8516507744789124,
+      "learning_rate": 8.083333333333334e-06,
+      "loss": 0.4064,
       "step": 3150
     },
     {
-      "epoch": 139.1304347826087,
-      "grad_norm": 0.7450351715087891,
-      "learning_rate": 8.005e-06,
-      "loss": 0.3923,
+      "epoch": 126.73267326732673,
+      "grad_norm": 0.6938304901123047,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.4057,
       "step": 3200
     },
     {
-      "epoch": 141.30434782608697,
-      "grad_norm": 1.05226731300354,
-      "learning_rate": 7.921666666666667e-06,
-      "loss": 0.3951,
+      "epoch": 128.7128712871287,
+      "grad_norm": 0.6889196038246155,
+      "learning_rate": 7.916666666666667e-06,
+      "loss": 0.4022,
       "step": 3250
     },
     {
-      "epoch": 143.47826086956522,
-      "grad_norm": 1.5119163990020752,
-      "learning_rate": 7.838333333333334e-06,
-      "loss": 0.3957,
+      "epoch": 130.69306930693068,
+      "grad_norm": 0.735680103302002,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 0.4019,
       "step": 3300
     },
     {
-      "epoch": 145.65217391304347,
-      "grad_norm": 1.0422383546829224,
-      "learning_rate": 7.755000000000001e-06,
-      "loss": 0.3942,
+      "epoch": 132.67326732673268,
+      "grad_norm": 1.0390517711639404,
+      "learning_rate": 7.75e-06,
+      "loss": 0.4014,
       "step": 3350
     },
     {
-      "epoch": 147.82608695652175,
-      "grad_norm": 1.2120935916900635,
-      "learning_rate": 7.671666666666668e-06,
-      "loss": 0.3946,
+      "epoch": 134.65346534653466,
+      "grad_norm": 0.807412326335907,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 0.3989,
       "step": 3400
     },
     {
-      "epoch": 150.0,
-      "grad_norm": 1.3127976655960083,
-      "learning_rate": 7.588333333333334e-06,
-      "loss": 0.3933,
+      "epoch": 136.63366336633663,
+      "grad_norm": 0.6928243637084961,
+      "learning_rate": 7.583333333333333e-06,
+      "loss": 0.4011,
       "step": 3450
     },
     {
-      "epoch": 152.17391304347825,
-      "grad_norm": 0.7704339027404785,
-      "learning_rate": 7.505e-06,
-      "loss": 0.3928,
+      "epoch": 138.6138613861386,
+      "grad_norm": 0.6894844770431519,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.4016,
       "step": 3500
     },
     {
-      "epoch": 152.17391304347825,
-      "eval_loss": 0.4132215976715088,
-      "eval_runtime": 6.6162,
-      "eval_samples_per_second": 24.032,
-      "eval_steps_per_second": 3.023,
+      "epoch": 138.6138613861386,
+      "eval_loss": 0.3931116759777069,
+      "eval_runtime": 7.7493,
+      "eval_samples_per_second": 23.228,
+      "eval_steps_per_second": 2.968,
       "step": 3500
     },
     {
-      "epoch": 154.34782608695653,
-      "grad_norm": 1.4827122688293457,
-      "learning_rate": 7.421666666666667e-06,
-      "loss": 0.3908,
+      "epoch": 140.59405940594058,
+      "grad_norm": 0.9431400895118713,
+      "learning_rate": 7.416666666666668e-06,
+      "loss": 0.4028,
       "step": 3550
     },
     {
-      "epoch": 156.52173913043478,
-      "grad_norm": 1.0464049577713013,
-      "learning_rate": 7.338333333333334e-06,
-      "loss": 0.3993,
+      "epoch": 142.5742574257426,
+      "grad_norm": 0.6703277230262756,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 0.3992,
       "step": 3600
     },
     {
-      "epoch": 158.69565217391303,
-      "grad_norm": 0.8279913663864136,
-      "learning_rate": 7.255000000000001e-06,
-      "loss": 0.3967,
+      "epoch": 144.55445544554456,
+      "grad_norm": 0.8334361910820007,
+      "learning_rate": 7.25e-06,
+      "loss": 0.3979,
       "step": 3650
     },
     {
-      "epoch": 160.8695652173913,
-      "grad_norm": 0.9175034165382385,
-      "learning_rate": 7.171666666666667e-06,
-      "loss": 0.3911,
+      "epoch": 146.53465346534654,
+      "grad_norm": 0.8266313672065735,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 0.3964,
       "step": 3700
     },
     {
-      "epoch": 163.04347826086956,
-      "grad_norm": 1.0080233812332153,
-      "learning_rate": 7.088333333333334e-06,
-      "loss": 0.3898,
+      "epoch": 148.5148514851485,
+      "grad_norm": 0.7688102126121521,
+      "learning_rate": 7.083333333333335e-06,
+      "loss": 0.3968,
       "step": 3750
     },
     {
-      "epoch": 165.2173913043478,
-      "grad_norm": 0.9944823980331421,
-      "learning_rate": 7.005000000000001e-06,
-      "loss": 0.3901,
+      "epoch": 150.4950495049505,
+      "grad_norm": 0.7516873478889465,
+      "learning_rate": 7e-06,
+      "loss": 0.3971,
       "step": 3800
     },
     {
-      "epoch": 167.3913043478261,
-      "grad_norm": 1.8222392797470093,
-      "learning_rate": 6.921666666666668e-06,
-      "loss": 0.3901,
+      "epoch": 152.47524752475246,
+      "grad_norm": 0.8034970760345459,
+      "learning_rate": 6.916666666666667e-06,
+      "loss": 0.3926,
       "step": 3850
     },
     {
-      "epoch": 169.56521739130434,
-      "grad_norm": 0.9506675004959106,
-      "learning_rate": 6.838333333333334e-06,
-      "loss": 0.3945,
+      "epoch": 154.45544554455446,
+      "grad_norm": 0.681676983833313,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 0.3975,
       "step": 3900
     },
     {
-      "epoch": 171.7391304347826,
-      "grad_norm": 4.512270450592041,
-      "learning_rate": 6.7550000000000005e-06,
-      "loss": 0.3913,
+      "epoch": 156.43564356435644,
+      "grad_norm": 0.6939652562141418,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 0.4005,
       "step": 3950
     },
     {
-      "epoch": 173.91304347826087,
-      "grad_norm": 0.74826580286026,
-      "learning_rate": 6.6716666666666674e-06,
-      "loss": 0.3914,
+      "epoch": 158.41584158415841,
+      "grad_norm": 0.6093599200248718,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.3937,
       "step": 4000
     },
     {
-      "epoch": 173.91304347826087,
-      "eval_loss": 0.4138501286506653,
-      "eval_runtime": 6.5399,
-      "eval_samples_per_second": 24.312,
-      "eval_steps_per_second": 3.058,
+      "epoch": 158.41584158415841,
+      "eval_loss": 0.39361944794654846,
+      "eval_runtime": 7.0786,
+      "eval_samples_per_second": 25.429,
+      "eval_steps_per_second": 3.249,
       "step": 4000
     },
     {
-      "epoch": 176.08695652173913,
-      "grad_norm": 1.4560046195983887,
-      "learning_rate": 6.588333333333334e-06,
-      "loss": 0.3903,
+      "epoch": 160.3960396039604,
+      "grad_norm": 0.6037696003913879,
+      "learning_rate": 6.5833333333333335e-06,
+      "loss": 0.3953,
       "step": 4050
     },
     {
-      "epoch": 178.2608695652174,
-      "grad_norm": 1.183271050453186,
-      "learning_rate": 6.505e-06,
-      "loss": 0.3901,
+      "epoch": 162.37623762376236,
+      "grad_norm": 0.6949267387390137,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.398,
       "step": 4100
     },
     {
-      "epoch": 180.43478260869566,
-      "grad_norm": 0.6697027683258057,
-      "learning_rate": 6.421666666666667e-06,
-      "loss": 0.399,
+      "epoch": 164.35643564356437,
+      "grad_norm": 0.7271093726158142,
+      "learning_rate": 6.418333333333334e-06,
+      "loss": 0.3945,
       "step": 4150
     },
     {
-      "epoch": 182.6086956521739,
-      "grad_norm": 0.96327143907547,
-      "learning_rate": 6.338333333333334e-06,
-      "loss": 0.3895,
+      "epoch": 166.33663366336634,
+      "grad_norm": 0.6959238052368164,
+      "learning_rate": 6.335e-06,
+      "loss": 0.3957,
       "step": 4200
     },
     {
-      "epoch": 184.7826086956522,
-      "grad_norm": 0.9833754301071167,
-      "learning_rate": 6.255e-06,
-      "loss": 0.3903,
+      "epoch": 168.31683168316832,
+      "grad_norm": 0.6987124681472778,
+      "learning_rate": 6.251666666666667e-06,
+      "loss": 0.3973,
       "step": 4250
     },
     {
-      "epoch": 186.95652173913044,
-      "grad_norm": 0.857953667640686,
-      "learning_rate": 6.171666666666667e-06,
-      "loss": 0.39,
+      "epoch": 170.2970297029703,
+      "grad_norm": 0.6888285875320435,
+      "learning_rate": 6.168333333333334e-06,
+      "loss": 0.3971,
       "step": 4300
     },
     {
-      "epoch": 189.1304347826087,
-      "grad_norm": 0.7506987452507019,
-      "learning_rate": 6.088333333333334e-06,
-      "loss": 0.394,
+      "epoch": 172.27722772277227,
+      "grad_norm": 0.6536576151847839,
+      "learning_rate": 6.085000000000001e-06,
+      "loss": 0.3939,
       "step": 4350
     },
     {
-      "epoch": 191.30434782608697,
-      "grad_norm": 0.8283782601356506,
-      "learning_rate": 6.005000000000001e-06,
-      "loss": 0.3869,
+      "epoch": 174.25742574257427,
+      "grad_norm": 0.6235485672950745,
+      "learning_rate": 6.001666666666667e-06,
+      "loss": 0.3949,
       "step": 4400
     },
     {
-      "epoch": 193.47826086956522,
-      "grad_norm": 1.0252714157104492,
-      "learning_rate": 5.921666666666667e-06,
-      "loss": 0.3967,
+      "epoch": 176.23762376237624,
+      "grad_norm": 0.7131723761558533,
+      "learning_rate": 5.918333333333334e-06,
+      "loss": 0.3944,
       "step": 4450
     },
     {
-      "epoch": 195.65217391304347,
-      "grad_norm": 0.9400416612625122,
-      "learning_rate": 5.838333333333334e-06,
-      "loss": 0.3916,
+      "epoch": 178.21782178217822,
+      "grad_norm": 0.5429028272628784,
+      "learning_rate": 5.835000000000001e-06,
+      "loss": 0.3953,
       "step": 4500
     },
     {
-      "epoch": 195.65217391304347,
-      "eval_loss": 0.41232919692993164,
-      "eval_runtime": 6.3912,
-      "eval_samples_per_second": 24.878,
-      "eval_steps_per_second": 3.129,
+      "epoch": 178.21782178217822,
+      "eval_loss": 0.3935137093067169,
+      "eval_runtime": 6.8281,
+      "eval_samples_per_second": 26.362,
+      "eval_steps_per_second": 3.368,
       "step": 4500
     },
     {
-      "epoch": 197.82608695652175,
-      "grad_norm": 0.9093385934829712,
-      "learning_rate": 5.755000000000001e-06,
-      "loss": 0.3894,
+      "epoch": 180.1980198019802,
+      "grad_norm": 0.6293529272079468,
+      "learning_rate": 5.751666666666668e-06,
+      "loss": 0.3923,
       "step": 4550
     },
     {
-      "epoch": 200.0,
-      "grad_norm": 1.078837513923645,
-      "learning_rate": 5.671666666666668e-06,
-      "loss": 0.3899,
+      "epoch": 182.17821782178217,
+      "grad_norm": 0.7379683256149292,
+      "learning_rate": 5.668333333333334e-06,
+      "loss": 0.393,
       "step": 4600
     },
     {
-      "epoch": 202.17391304347825,
-      "grad_norm": 0.7748771905899048,
-      "learning_rate": 5.588333333333334e-06,
-      "loss": 0.3943,
+      "epoch": 184.15841584158414,
+      "grad_norm": 0.6412554979324341,
+      "learning_rate": 5.585000000000001e-06,
+      "loss": 0.3915,
       "step": 4650
     },
     {
-      "epoch": 204.34782608695653,
-      "grad_norm": 1.1522585153579712,
-      "learning_rate": 5.505000000000001e-06,
-      "loss": 0.389,
+      "epoch": 186.13861386138615,
+      "grad_norm": 0.7281537652015686,
+      "learning_rate": 5.501666666666668e-06,
+      "loss": 0.3928,
       "step": 4700
     },
     {
-      "epoch": 206.52173913043478,
-      "grad_norm": 1.6818900108337402,
-      "learning_rate": 5.4216666666666676e-06,
-      "loss": 0.3961,
+      "epoch": 188.11881188118812,
+      "grad_norm": 0.640044093132019,
+      "learning_rate": 5.418333333333333e-06,
+      "loss": 0.3935,
       "step": 4750
     },
     {
-      "epoch": 208.69565217391303,
-      "grad_norm": 0.9630489349365234,
-      "learning_rate": 5.3383333333333345e-06,
-      "loss": 0.3926,
+      "epoch": 190.0990099009901,
+      "grad_norm": 0.6560197472572327,
+      "learning_rate": 5.335000000000001e-06,
+      "loss": 0.3922,
       "step": 4800
     },
     {
-      "epoch": 210.8695652173913,
-      "grad_norm": 0.915107250213623,
-      "learning_rate": 5.2550000000000005e-06,
-      "loss": 0.3843,
+      "epoch": 192.07920792079207,
+      "grad_norm": 0.5895935893058777,
+      "learning_rate": 5.2516666666666675e-06,
+      "loss": 0.3939,
       "step": 4850
     },
     {
-      "epoch": 213.04347826086956,
-      "grad_norm": 0.7301034331321716,
-      "learning_rate": 5.171666666666667e-06,
-      "loss": 0.3929,
+      "epoch": 194.05940594059405,
+      "grad_norm": 0.7578818202018738,
+      "learning_rate": 5.168333333333334e-06,
+      "loss": 0.3939,
       "step": 4900
     },
     {
-      "epoch": 215.2173913043478,
-      "grad_norm": 0.8523257970809937,
-      "learning_rate": 5.088333333333334e-06,
-      "loss": 0.3873,
+      "epoch": 196.03960396039605,
+      "grad_norm": 0.6305426955223083,
+      "learning_rate": 5.085e-06,
+      "loss": 0.3917,
       "step": 4950
     },
     {
-      "epoch": 217.3913043478261,
-      "grad_norm": 0.7538474798202515,
-      "learning_rate": 5.0049999999999995e-06,
-      "loss": 0.3919,
+      "epoch": 198.01980198019803,
+      "grad_norm": 0.8592019081115723,
+      "learning_rate": 5.0016666666666665e-06,
+      "loss": 0.3933,
       "step": 5000
     },
     {
-      "epoch": 217.3913043478261,
-      "eval_loss": 0.4122372269630432,
-      "eval_runtime": 6.5201,
-      "eval_samples_per_second": 24.386,
-      "eval_steps_per_second": 3.067,
+      "epoch": 198.01980198019803,
+      "eval_loss": 0.39355847239494324,
+      "eval_runtime": 7.4596,
+      "eval_samples_per_second": 24.13,
+      "eval_steps_per_second": 3.083,
       "step": 5000
     },
     {
-      "epoch": 219.56521739130434,
-      "grad_norm": 1.006597638130188,
-      "learning_rate": 4.921666666666666e-06,
-      "loss": 0.3844,
+      "epoch": 200.0,
+      "grad_norm": 0.9197565317153931,
+      "learning_rate": 4.918333333333334e-06,
+      "loss": 0.3947,
       "step": 5050
     },
     {
-      "epoch": 221.7391304347826,
-      "grad_norm": 0.6486532092094421,
-      "learning_rate": 4.838333333333334e-06,
-      "loss": 0.3853,
+      "epoch": 201.98019801980197,
+      "grad_norm": 0.675254225730896,
+      "learning_rate": 4.835e-06,
+      "loss": 0.3921,
       "step": 5100
     },
     {
-      "epoch": 223.91304347826087,
-      "grad_norm": 0.7955175638198853,
-      "learning_rate": 4.755e-06,
-      "loss": 0.3853,
+      "epoch": 203.96039603960395,
+      "grad_norm": 0.7896837592124939,
+      "learning_rate": 4.751666666666667e-06,
+      "loss": 0.3937,
       "step": 5150
     },
     {
-      "epoch": 226.08695652173913,
-      "grad_norm": 0.8011429309844971,
-      "learning_rate": 4.671666666666667e-06,
-      "loss": 0.3888,
+      "epoch": 205.94059405940595,
+      "grad_norm": 0.6978718042373657,
+      "learning_rate": 4.668333333333333e-06,
+      "loss": 0.3912,
       "step": 5200
     },
     {
-      "epoch": 228.2608695652174,
-      "grad_norm": 0.7421734929084778,
-      "learning_rate": 4.588333333333333e-06,
-      "loss": 0.3945,
+      "epoch": 207.92079207920793,
+      "grad_norm": 0.6968542337417603,
+      "learning_rate": 4.585e-06,
+      "loss": 0.3902,
       "step": 5250
     },
     {
-      "epoch": 230.43478260869566,
-      "grad_norm": 1.5567547082901,
-      "learning_rate": 4.505e-06,
-      "loss": 0.3916,
+      "epoch": 209.9009900990099,
+      "grad_norm": 0.6489793062210083,
+      "learning_rate": 4.501666666666667e-06,
+      "loss": 0.3911,
       "step": 5300
     },
     {
-      "epoch": 232.6086956521739,
-      "grad_norm": 0.8348387479782104,
-      "learning_rate": 4.421666666666667e-06,
-      "loss": 0.3887,
+      "epoch": 211.88118811881188,
+      "grad_norm": 0.6600732803344727,
+      "learning_rate": 4.418333333333334e-06,
+      "loss": 0.3912,
       "step": 5350
     },
     {
-      "epoch": 234.7826086956522,
-      "grad_norm": 0.965840756893158,
-      "learning_rate": 4.338333333333334e-06,
-      "loss": 0.3907,
+      "epoch": 213.86138613861385,
+      "grad_norm": 0.6637029051780701,
+      "learning_rate": 4.335e-06,
+      "loss": 0.3903,
       "step": 5400
     },
     {
-      "epoch": 236.95652173913044,
-      "grad_norm": 1.0057460069656372,
-      "learning_rate": 4.255e-06,
-      "loss": 0.4004,
+      "epoch": 215.84158415841586,
+      "grad_norm": 0.9407865405082703,
+      "learning_rate": 4.251666666666667e-06,
+      "loss": 0.3913,
       "step": 5450
     },
     {
-      "epoch": 239.1304347826087,
-      "grad_norm": 1.2100751399993896,
-      "learning_rate": 4.171666666666667e-06,
-      "loss": 0.3844,
+      "epoch": 217.82178217821783,
+      "grad_norm": 0.674074649810791,
+      "learning_rate": 4.168333333333334e-06,
+      "loss": 0.3932,
       "step": 5500
     },
     {
-      "epoch": 239.1304347826087,
-      "eval_loss": 0.41257867217063904,
-      "eval_runtime": 6.47,
-      "eval_samples_per_second": 24.575,
-      "eval_steps_per_second": 3.091,
+      "epoch": 217.82178217821783,
+      "eval_loss": 0.3936135172843933,
+      "eval_runtime": 7.411,
+      "eval_samples_per_second": 24.288,
+      "eval_steps_per_second": 3.103,
       "step": 5500
     },
     {
-      "epoch": 241.30434782608697,
-      "grad_norm": 0.7117742896080017,
-      "learning_rate": 4.088333333333334e-06,
-      "loss": 0.3845,
+      "epoch": 219.8019801980198,
+      "grad_norm": 0.5686890482902527,
+      "learning_rate": 4.085e-06,
+      "loss": 0.391,
       "step": 5550
     },
     {
-      "epoch": 243.47826086956522,
-      "grad_norm": 0.6881281733512878,
-      "learning_rate": 4.005000000000001e-06,
-      "loss": 0.3858,
+      "epoch": 221.78217821782178,
+      "grad_norm": 0.5418526530265808,
+      "learning_rate": 4.001666666666667e-06,
+      "loss": 0.3863,
       "step": 5600
     },
     {
-      "epoch": 245.65217391304347,
-      "grad_norm": 1.3413081169128418,
-      "learning_rate": 3.921666666666667e-06,
-      "loss": 0.3842,
+      "epoch": 223.76237623762376,
+      "grad_norm": 0.7057704925537109,
+      "learning_rate": 3.918333333333334e-06,
+      "loss": 0.394,
       "step": 5650
     },
     {
-      "epoch": 247.82608695652175,
-      "grad_norm": 0.7762352228164673,
-      "learning_rate": 3.8383333333333336e-06,
-      "loss": 0.3839,
+      "epoch": 225.74257425742573,
+      "grad_norm": 0.6317301392555237,
+      "learning_rate": 3.8350000000000006e-06,
+      "loss": 0.3906,
       "step": 5700
     },
     {
-      "epoch": 250.0,
-      "grad_norm": 1.2861838340759277,
-      "learning_rate": 3.7550000000000005e-06,
-      "loss": 0.3836,
+      "epoch": 227.72277227722773,
+      "grad_norm": 0.6327551007270813,
+      "learning_rate": 3.7516666666666666e-06,
+      "loss": 0.3902,
       "step": 5750
     },
     {
-      "epoch": 252.17391304347825,
-      "grad_norm": 1.12875497341156,
-      "learning_rate": 3.6716666666666665e-06,
-      "loss": 0.3828,
+      "epoch": 229.7029702970297,
+      "grad_norm": 0.6199086904525757,
+      "learning_rate": 3.668333333333334e-06,
+      "loss": 0.3912,
       "step": 5800
     },
     {
-      "epoch": 254.34782608695653,
-      "grad_norm": 0.9641019105911255,
-      "learning_rate": 3.588333333333334e-06,
-      "loss": 0.3815,
+      "epoch": 231.68316831683168,
+      "grad_norm": 0.6341702938079834,
+      "learning_rate": 3.585e-06,
+      "loss": 0.391,
       "step": 5850
     },
     {
-      "epoch": 256.5217391304348,
-      "grad_norm": 0.7063067555427551,
-      "learning_rate": 3.505e-06,
-      "loss": 0.3838,
+      "epoch": 233.66336633663366,
+      "grad_norm": 0.7226659655570984,
+      "learning_rate": 3.5016666666666673e-06,
+      "loss": 0.3908,
       "step": 5900
     },
     {
-      "epoch": 258.69565217391306,
-      "grad_norm": 1.0992182493209839,
-      "learning_rate": 3.4216666666666672e-06,
-      "loss": 0.3824,
+      "epoch": 235.64356435643563,
+      "grad_norm": 0.6628900170326233,
+      "learning_rate": 3.4183333333333334e-06,
+      "loss": 0.3899,
       "step": 5950
     },
     {
-      "epoch": 260.8695652173913,
-      "grad_norm": 0.6153731942176819,
-      "learning_rate": 3.3383333333333333e-06,
-      "loss": 0.3908,
+      "epoch": 237.62376237623764,
+      "grad_norm": 0.6972154378890991,
+      "learning_rate": 3.3350000000000003e-06,
+      "loss": 0.3884,
       "step": 6000
     },
     {
-      "epoch": 260.8695652173913,
-      "eval_loss": 0.4136714041233063,
-      "eval_runtime": 6.4416,
-      "eval_samples_per_second": 24.683,
-      "eval_steps_per_second": 3.105,
+      "epoch": 237.62376237623764,
+      "eval_loss": 0.3934413194656372,
+      "eval_runtime": 7.3097,
+      "eval_samples_per_second": 24.625,
+      "eval_steps_per_second": 3.146,
       "step": 6000
     },
     {
-      "epoch": 263.04347826086956,
-      "grad_norm": 0.8322598934173584,
-      "learning_rate": 3.255e-06,
-      "loss": 0.3918,
+      "epoch": 239.6039603960396,
+      "grad_norm": 0.7429386377334595,
+      "learning_rate": 3.2516666666666667e-06,
+      "loss": 0.3902,
       "step": 6050
     },
     {
-      "epoch": 265.2173913043478,
-      "grad_norm": 0.7336969971656799,
-      "learning_rate": 3.1716666666666667e-06,
-      "loss": 0.3908,
+      "epoch": 241.58415841584159,
+      "grad_norm": 0.6832717061042786,
+      "learning_rate": 3.1683333333333336e-06,
+      "loss": 0.3896,
       "step": 6100
     },
     {
-      "epoch": 267.39130434782606,
-      "grad_norm": 0.7218737006187439,
-      "learning_rate": 3.0883333333333336e-06,
-      "loss": 0.3848,
+      "epoch": 243.56435643564356,
+      "grad_norm": 0.7591320872306824,
+      "learning_rate": 3.085e-06,
+      "loss": 0.3905,
       "step": 6150
     },
     {
-      "epoch": 269.5652173913044,
-      "grad_norm": 1.2083243131637573,
-      "learning_rate": 3.005e-06,
-      "loss": 0.3868,
+      "epoch": 245.54455445544554,
+      "grad_norm": 0.5904133319854736,
+      "learning_rate": 3.001666666666667e-06,
+      "loss": 0.3876,
       "step": 6200
     },
     {
-      "epoch": 271.7391304347826,
-      "grad_norm": 0.9182942509651184,
-      "learning_rate": 2.921666666666667e-06,
-      "loss": 0.3895,
+      "epoch": 247.52475247524754,
+      "grad_norm": 0.6338399052619934,
+      "learning_rate": 2.9183333333333335e-06,
+      "loss": 0.3891,
       "step": 6250
     },
     {
-      "epoch": 273.9130434782609,
-      "grad_norm": 0.8747413158416748,
-      "learning_rate": 2.8383333333333334e-06,
-      "loss": 0.3842,
+      "epoch": 249.5049504950495,
+      "grad_norm": 0.6359020471572876,
+      "learning_rate": 2.835e-06,
+      "loss": 0.3905,
       "step": 6300
     },
     {
-      "epoch": 276.0869565217391,
-      "grad_norm": 0.8834450840950012,
-      "learning_rate": 2.7550000000000003e-06,
-      "loss": 0.3879,
+      "epoch": 251.4851485148515,
+      "grad_norm": 0.6320521831512451,
+      "learning_rate": 2.751666666666667e-06,
+      "loss": 0.3885,
       "step": 6350
     },
     {
-      "epoch": 278.2608695652174,
-      "grad_norm": 1.0217564105987549,
-      "learning_rate": 2.6716666666666668e-06,
-      "loss": 0.3892,
+      "epoch": 253.46534653465346,
+      "grad_norm": 0.6006565690040588,
+      "learning_rate": 2.6683333333333333e-06,
+      "loss": 0.3861,
       "step": 6400
     },
     {
-      "epoch": 280.4347826086956,
-      "grad_norm": 0.6895825862884521,
-      "learning_rate": 2.5883333333333337e-06,
-      "loss": 0.389,
+      "epoch": 255.44554455445544,
+      "grad_norm": 0.5949265360832214,
+      "learning_rate": 2.5866666666666667e-06,
+      "loss": 0.3898,
       "step": 6450
     },
     {
-      "epoch": 282.60869565217394,
-      "grad_norm": 0.6623283624649048,
-      "learning_rate": 2.505e-06,
-      "loss": 0.3854,
+      "epoch": 257.4257425742574,
+      "grad_norm": 0.6407713294029236,
+      "learning_rate": 2.5033333333333336e-06,
+      "loss": 0.3889,
       "step": 6500
     },
     {
-      "epoch": 282.60869565217394,
-      "eval_loss": 0.413311630487442,
-      "eval_runtime": 6.7085,
-      "eval_samples_per_second": 23.701,
-      "eval_steps_per_second": 2.981,
+      "epoch": 257.4257425742574,
+      "eval_loss": 0.3926040828227997,
+      "eval_runtime": 7.8449,
+      "eval_samples_per_second": 22.945,
+      "eval_steps_per_second": 2.932,
       "step": 6500
     },
     {
-      "epoch": 284.7826086956522,
-      "grad_norm": 0.7647244930267334,
-      "learning_rate": 2.421666666666667e-06,
-      "loss": 0.384,
+      "epoch": 259.4059405940594,
+      "grad_norm": 0.6776269674301147,
+      "learning_rate": 2.42e-06,
+      "loss": 0.3863,
       "step": 6550
     },
     {
-      "epoch": 286.95652173913044,
-      "grad_norm": 0.7481832504272461,
-      "learning_rate": 2.3383333333333335e-06,
-      "loss": 0.3897,
+      "epoch": 261.38613861386136,
+      "grad_norm": 0.7183690667152405,
+      "learning_rate": 2.3366666666666666e-06,
+      "loss": 0.388,
       "step": 6600
     },
     {
-      "epoch": 289.1304347826087,
-      "grad_norm": 0.9159369468688965,
-      "learning_rate": 2.2550000000000004e-06,
-      "loss": 0.3852,
+      "epoch": 263.36633663366337,
+      "grad_norm": 0.683351993560791,
+      "learning_rate": 2.2533333333333335e-06,
+      "loss": 0.3861,
       "step": 6650
     },
     {
-      "epoch": 291.30434782608694,
-      "grad_norm": 0.5772305727005005,
-      "learning_rate": 2.171666666666667e-06,
-      "loss": 0.3864,
+      "epoch": 265.34653465346537,
+      "grad_norm": 0.5622109770774841,
+      "learning_rate": 2.17e-06,
+      "loss": 0.3872,
       "step": 6700
     },
     {
-      "epoch": 293.4782608695652,
-      "grad_norm": 0.8706581592559814,
-      "learning_rate": 2.088333333333334e-06,
-      "loss": 0.3818,
+      "epoch": 267.3267326732673,
+      "grad_norm": 0.6307435035705566,
+      "learning_rate": 2.086666666666667e-06,
+      "loss": 0.3851,
       "step": 6750
     },
     {
-      "epoch": 295.6521739130435,
-      "grad_norm": 0.6359920501708984,
-      "learning_rate": 2.0050000000000003e-06,
-      "loss": 0.3866,
+      "epoch": 269.3069306930693,
+      "grad_norm": 0.5858733654022217,
+      "learning_rate": 2.0033333333333334e-06,
+      "loss": 0.3847,
       "step": 6800
     },
     {
-      "epoch": 297.82608695652175,
-      "grad_norm": 0.6971527934074402,
-      "learning_rate": 1.9216666666666668e-06,
-      "loss": 0.3835,
+      "epoch": 271.28712871287127,
+      "grad_norm": 0.7978197336196899,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.3876,
       "step": 6850
     },
     {
-      "epoch": 300.0,
-      "grad_norm": 1.396098017692566,
-      "learning_rate": 1.8383333333333334e-06,
-      "loss": 0.3878,
+      "epoch": 273.26732673267327,
+      "grad_norm": 0.6843935251235962,
+      "learning_rate": 1.836666666666667e-06,
+      "loss": 0.3871,
       "step": 6900
     },
     {
-      "epoch": 302.17391304347825,
-      "grad_norm": 0.6345895528793335,
-      "learning_rate": 1.7550000000000001e-06,
-      "loss": 0.382,
+      "epoch": 275.2475247524753,
+      "grad_norm": 0.8366850018501282,
+      "learning_rate": 1.7533333333333336e-06,
+      "loss": 0.3909,
       "step": 6950
     },
     {
-      "epoch": 304.3478260869565,
-      "grad_norm": 1.3194807767868042,
-      "learning_rate": 1.6716666666666666e-06,
-      "loss": 0.3885,
+      "epoch": 277.2277227722772,
+      "grad_norm": 0.584172248840332,
+      "learning_rate": 1.6700000000000003e-06,
+      "loss": 0.3855,
       "step": 7000
     },
     {
-      "epoch": 304.3478260869565,
-      "eval_loss": 0.4140428602695465,
-      "eval_runtime": 6.4587,
-      "eval_samples_per_second": 24.618,
-      "eval_steps_per_second": 3.097,
+      "epoch": 277.2277227722772,
+      "eval_loss": 0.39400026202201843,
+      "eval_runtime": 6.8593,
+      "eval_samples_per_second": 26.242,
+      "eval_steps_per_second": 3.353,
       "step": 7000
     },
     {
-      "epoch": 306.5217391304348,
-      "grad_norm": 0.8248822689056396,
-      "learning_rate": 1.5883333333333333e-06,
-      "loss": 0.3847,
+      "epoch": 279.2079207920792,
+      "grad_norm": 0.7005228996276855,
+      "learning_rate": 1.586666666666667e-06,
+      "loss": 0.3858,
       "step": 7050
     },
     {
-      "epoch": 308.69565217391306,
-      "grad_norm": 0.635430097579956,
-      "learning_rate": 1.505e-06,
-      "loss": 0.3894,
+      "epoch": 281.18811881188117,
+      "grad_norm": 0.5294095277786255,
+      "learning_rate": 1.5033333333333337e-06,
+      "loss": 0.3845,
       "step": 7100
     },
     {
-      "epoch": 310.8695652173913,
-      "grad_norm": 0.8129938244819641,
-      "learning_rate": 1.4216666666666667e-06,
-      "loss": 0.3862,
+      "epoch": 283.16831683168317,
+      "grad_norm": 0.5473897457122803,
+      "learning_rate": 1.42e-06,
+      "loss": 0.3846,
       "step": 7150
     },
     {
-      "epoch": 313.04347826086956,
-      "grad_norm": 0.8096001148223877,
-      "learning_rate": 1.3383333333333334e-06,
-      "loss": 0.3859,
+      "epoch": 285.1485148514852,
+      "grad_norm": 0.6222283840179443,
+      "learning_rate": 1.3366666666666666e-06,
+      "loss": 0.3849,
       "step": 7200
     },
     {
-      "epoch": 315.2173913043478,
-      "grad_norm": 0.7998645901679993,
-      "learning_rate": 1.255e-06,
-      "loss": 0.3826,
+      "epoch": 287.1287128712871,
+      "grad_norm": 0.6396065354347229,
+      "learning_rate": 1.2533333333333333e-06,
+      "loss": 0.3878,
       "step": 7250
     },
     {
-      "epoch": 317.39130434782606,
-      "grad_norm": 0.6718857288360596,
-      "learning_rate": 1.1716666666666667e-06,
-      "loss": 0.3896,
+      "epoch": 289.1089108910891,
+      "grad_norm": 0.6903032064437866,
+      "learning_rate": 1.1700000000000002e-06,
+      "loss": 0.3865,
       "step": 7300
     },
     {
-      "epoch": 319.5652173913044,
-      "grad_norm": 0.5565139055252075,
-      "learning_rate": 1.0883333333333334e-06,
-      "loss": 0.3852,
+      "epoch": 291.08910891089107,
+      "grad_norm": 0.6112281680107117,
+      "learning_rate": 1.0866666666666667e-06,
+      "loss": 0.3888,
       "step": 7350
     },
     {
-      "epoch": 321.7391304347826,
-      "grad_norm": 0.7727565765380859,
-      "learning_rate": 1.0050000000000001e-06,
-      "loss": 0.3866,
+      "epoch": 293.0693069306931,
+      "grad_norm": 0.6655446887016296,
+      "learning_rate": 1.0033333333333334e-06,
+      "loss": 0.3844,
       "step": 7400
     },
     {
-      "epoch": 323.9130434782609,
-      "grad_norm": 0.7541866302490234,
-      "learning_rate": 9.216666666666667e-07,
-      "loss": 0.3818,
+      "epoch": 295.0495049504951,
+      "grad_norm": 0.5784509181976318,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 0.3862,
       "step": 7450
     },
     {
-      "epoch": 326.0869565217391,
-      "grad_norm": 0.7003363966941833,
-      "learning_rate": 8.383333333333334e-07,
-      "loss": 0.3849,
+      "epoch": 297.029702970297,
+      "grad_norm": 0.6216564774513245,
+      "learning_rate": 8.366666666666668e-07,
+      "loss": 0.3892,
       "step": 7500
     },
     {
-      "epoch": 326.0869565217391,
-      "eval_loss": 0.41430386900901794,
-      "eval_runtime": 6.7779,
-      "eval_samples_per_second": 23.458,
-      "eval_steps_per_second": 2.951,
+      "epoch": 297.029702970297,
+      "eval_loss": 0.3934391140937805,
+      "eval_runtime": 7.8487,
+      "eval_samples_per_second": 22.934,
+      "eval_steps_per_second": 2.93,
       "step": 7500
     },
     {
-      "epoch": 328.2608695652174,
-      "grad_norm": 0.5425256490707397,
-      "learning_rate": 7.550000000000001e-07,
-      "loss": 0.39,
+      "epoch": 299.009900990099,
+      "grad_norm": 0.9473925232887268,
+      "learning_rate": 7.533333333333335e-07,
+      "loss": 0.3873,
       "step": 7550
     },
     {
-      "epoch": 330.4347826086956,
-      "grad_norm": 0.7711132168769836,
-      "learning_rate": 6.716666666666668e-07,
+      "epoch": 300.990099009901,
+      "grad_norm": 0.5682732462882996,
+      "learning_rate": 6.7e-07,
       "loss": 0.3874,
       "step": 7600
     },
     {
-      "epoch": 332.60869565217394,
-      "grad_norm": 0.7697465419769287,
-      "learning_rate": 5.883333333333333e-07,
-      "loss": 0.3878,
+      "epoch": 302.970297029703,
+      "grad_norm": 0.787708580493927,
+      "learning_rate": 5.866666666666667e-07,
+      "loss": 0.3892,
       "step": 7650
     },
     {
-      "epoch": 334.7826086956522,
-      "grad_norm": 0.679308295249939,
-      "learning_rate": 5.05e-07,
-      "loss": 0.3888,
+      "epoch": 304.9504950495049,
+      "grad_norm": 0.7085989713668823,
+      "learning_rate": 5.033333333333334e-07,
+      "loss": 0.3851,
       "step": 7700
     },
     {
-      "epoch": 336.95652173913044,
-      "grad_norm": 0.7001920938491821,
-      "learning_rate": 4.2166666666666667e-07,
-      "loss": 0.3809,
+      "epoch": 306.9306930693069,
+      "grad_norm": 0.6272616386413574,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 0.3891,
       "step": 7750
     },
     {
-      "epoch": 339.1304347826087,
-      "grad_norm": 0.7666972875595093,
-      "learning_rate": 3.3833333333333336e-07,
-      "loss": 0.3788,
+      "epoch": 308.91089108910893,
+      "grad_norm": 0.6887345314025879,
+      "learning_rate": 3.366666666666667e-07,
+      "loss": 0.3874,
       "step": 7800
     },
     {
-      "epoch": 341.30434782608694,
-      "grad_norm": 0.8930239677429199,
-      "learning_rate": 2.55e-07,
-      "loss": 0.3841,
+      "epoch": 310.8910891089109,
+      "grad_norm": 0.6206291913986206,
+      "learning_rate": 2.533333333333333e-07,
+      "loss": 0.3853,
       "step": 7850
     },
     {
-      "epoch": 343.4782608695652,
-      "grad_norm": 0.7171005606651306,
-      "learning_rate": 1.7166666666666668e-07,
-      "loss": 0.386,
+      "epoch": 312.8712871287129,
+      "grad_norm": 0.6549195051193237,
+      "learning_rate": 1.7000000000000001e-07,
+      "loss": 0.3867,
       "step": 7900
     },
     {
-      "epoch": 345.6521739130435,
-      "grad_norm": 1.4843720197677612,
-      "learning_rate": 8.833333333333334e-08,
-      "loss": 0.3846,
+      "epoch": 314.8514851485148,
+      "grad_norm": 0.7210911512374878,
+      "learning_rate": 8.666666666666668e-08,
+      "loss": 0.3845,
       "step": 7950
     },
     {
-      "epoch": 347.82608695652175,
-      "grad_norm": 0.7774099707603455,
-      "learning_rate": 5e-09,
-      "loss": 0.386,
+      "epoch": 316.83168316831683,
+      "grad_norm": 0.6023419499397278,
+      "learning_rate": 3.3333333333333334e-09,
+      "loss": 0.3899,
       "step": 8000
     },
     {
-      "epoch": 347.82608695652175,
-      "eval_loss": 0.4134390950202942,
-      "eval_runtime": 6.5968,
-      "eval_samples_per_second": 24.103,
-      "eval_steps_per_second": 3.032,
+      "epoch": 316.83168316831683,
+      "eval_loss": 0.39394611120224,
+      "eval_runtime": 7.5969,
+      "eval_samples_per_second": 23.694,
+      "eval_steps_per_second": 3.028,
       "step": 8000
     }
   ],
   "logging_steps": 50,
   "max_steps": 8000,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 348,
+  "num_train_epochs": 320,
   "save_steps": 500,
   "stateful_callbacks": {
     "TrainerControl": {
@@ -1274,8 +1274,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 1.1839551420994704e+17,
-  "train_batch_size": 64,
+  "total_flos": 8.643923525044128e+16,
+  "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null
 }

checkpoint-8000/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f95de2b194fe7f16ec10fbf1b93d83e04c25498d479793b9a5748c1d338bfbd0
+oid sha256:7e97537a0fb344678f807f30d02bd561fb4b5c2162a68338251550dd3ef50d20
 size 5304